mom_coms mom_coms::assignment(=) mom_coms::efp_sum_across_pes mom_coms::efp_type mom_coms::operator(+) mom_coms::operator(-) mom_coms::reproducing_sum mom_coms::reproducing_sum_efp integer(kind=8), parameter integer(kind=8), parameter mom_coms::prec prec =2_8**46 The precision of each integer. carry_overflow increment_ints real_to_ints regularize_ints real, parameter real, parameter mom_coms::r_prec r_prec =2.0**46 A real version of prec. real, parameter real, parameter mom_coms::i_prec i_prec =1.0/(2.0**46) The inverse of prec. carry_overflow regularize_ints integer, parameter integer, parameter mom_coms::max_count_prec max_count_prec =2**(63-46)-1 The number of values that can be added together with the current value of prec before there will be roundoff problems. efp_list_sum_across_pes efp_val_sum_across_pes reproducing_efp_sum_2d reproducing_sum_2d reproducing_sum_3d integer, parameter integer, parameter mom_coms::ni ni =6 The number of long integers to use to represent a real number. carry_overflow efp_assign efp_list_sum_across_pes efp_minus efp_val_sum_across_pes increment_ints increment_ints_faster ints_to_real real_to_ints regularize_ints reproducing_efp_sum_2d reproducing_sum_2d reproducing_sum_3d real, dimension(ni), parameter real, dimension(ni), parameter mom_coms::pr pr = (/ r_prec**2, r_prec, 1.0, 1.0/r_prec, 1.0/r_prec**2, 1.0/r_prec**3 /) An array of the real precision of each of the integers. increment_ints_faster ints_to_real real_to_ints reproducing_efp_sum_2d reproducing_sum_3d real, dimension(ni), parameter real, dimension(ni), parameter mom_coms::i_pr i_pr = (/ 1.0/r_prec**2, 1.0/r_prec, 1.0, r_prec, r_prec**2, r_prec**3 /) An array of the inverse of the real precision of each of the integers. increment_ints_faster real_to_ints reproducing_efp_sum_2d real, parameter real, parameter mom_coms::max_efp_float max_efp_float = pr(1) * (2.**63 - 1.) The largest float with an EFP representation. NOTE: Only the first bin can exceed precision, but is bounded by the largest signed integer. increment_ints_faster logical logical mom_coms::overflow_error overflow_error = .false. This becomes true if an overflow is encountered. carry_overflow efp_list_sum_across_pes efp_val_sum_across_pes increment_ints increment_ints_faster query_efp_overflow_error reproducing_efp_sum_2d reproducing_sum_2d reproducing_sum_3d reset_efp_overflow_error logical logical mom_coms::nan_error nan_error = .false. This becomes true if a NaN is encountered. increment_ints_faster real_to_ints reproducing_efp_sum_2d reproducing_sum_3d logical logical mom_coms::debug debug = .false. Making this true enables debugging output. reproducing_sum_2d reproducing_sum_3d type(efp_type) function type(efp_type) function mom_coms::reproducing_efp_sum_2d (array, isr, ier, jsr, jer, overflow_check, err, only_on_PE) reproducing_efp_sum_2d array array isr isr ier ier jsr jsr jer jer overflow_check overflow_check err err only_on_PE only_on_PE This subroutine uses a conversion to an integer representation of real numbers to give an order-invariant sum of distributed 2-D arrays that reproduces across domain decomposition, with the result returned as an extended fixed point type that can be converted back to a real number using EFP_to_real. This technique is described in Hallberg & Adcroft, 2014, Parallel Computing, doi:10.1016/j.parco.2014.04.007. array The array to be summed isr The starting i-index of the sum, noting that the array indices starts at 1 ier The ending i-index of the sum, noting that the array indices starts at 1 jsr The starting j-index of the sum, noting that the array indices starts at 1 jer The ending j-index of the sum, noting that the array indices starts at 1 overflow_check If present and false, disable checking for overflows in incremental results. This can speed up calculations if the number of values being summed is small enough err If present, return an error code instead of triggering any fatal errors directly from this routine. only_on_pe If present and true, do not do the sum across processors, only reporting the local sum The result in extended fixed point format carry_overflow i_pr increment_ints increment_ints_faster max_count_prec mom_error_handler::mom_error nan_error ni overflow_error pr real_to_ints regularize_ints reproducing_sum_2d real function real function mom_coms::reproducing_sum_2d (array, isr, ier, jsr, jer, EFP_sum, reproducing, overflow_check, err, only_on_PE) reproducing_sum_2d array array isr isr ier ier jsr jsr jer jer EFP_sum EFP_sum reproducing reproducing overflow_check overflow_check err err only_on_PE only_on_PE This subroutine uses a conversion to an integer representation of real numbers to give an order-invariant sum of distributed 2-D arrays that reproduces across domain decomposition. This technique is described in Hallberg & Adcroft, 2014, Parallel Computing, doi:10.1016/j.parco.2014.04.007. array The array to be summed isr The starting i-index of the sum, noting that the array indices starts at 1 ier The ending i-index of the sum, noting that the array indices starts at 1 jsr The starting j-index of the sum, noting that the array indices starts at 1 jer The ending j-index of the sum, noting that the array indices starts at 1 efp_sum The result in extended fixed point format reproducing If present and false, do the sum using the naive non-reproducing approach overflow_check If present and false, disable checking for overflows in incremental results. This can speed up calculations if the number of values being summed is small enough err If present, return an error code instead of triggering any fatal errors directly from this routine. only_on_pe If present and true, do not do the sum across processors, only reporting the local sum Result debug ints_to_real max_count_prec mom_error_handler::mom_error mom_error_handler::mom_mesg ni overflow_error real_to_ints reproducing_efp_sum_2d real function real function mom_coms::reproducing_sum_3d (array, isr, ier, jsr, jer, sums, EFP_sum, EFP_lay_sums, err, only_on_PE) reproducing_sum_3d array array isr isr ier ier jsr jsr jer jer sums sums EFP_sum EFP_sum EFP_lay_sums EFP_lay_sums err err only_on_PE only_on_PE This subroutine uses a conversion to an integer representation of real numbers to give an order-invariant sum of distributed 3-D arrays that reproduces across domain decomposition. This technique is described in Hallberg & Adcroft, 2014, Parallel Computing, doi:10.1016/j.parco.2014.04.007. array The array to be summed isr The starting i-index of the sum, noting that the array indices starts at 1 ier The ending i-index of the sum, noting that the array indices starts at 1 jsr The starting j-index of the sum, noting that the array indices starts at 1 jer The ending j-index of the sum, noting that the array indices starts at 1 sums The sums by vertical layer efp_sum The result in extended fixed point format efp_lay_sums The sums by vertical layer in EFP format err If present, return an error code instead of triggering any fatal errors directly from this routine. only_on_pe If present and true, do not do the sum across processors, only reporting the local sum Result carry_overflow debug increment_ints increment_ints_faster ints_to_real max_count_prec mom_error_handler::mom_error mom_error_handler::mom_mesg nan_error ni overflow_error pr real_to_ints regularize_ints integer(kind=8) function, dimension(ni) integer(kind=8) function, dimension(ni) mom_coms::real_to_ints (r, prec_error, overflow) real_to_ints r r prec_error prec_error overflow overflow Convert a real number into the array of integers constitute its extended-fixed-point representation. r The real number being converted prec_error The PE-count dependent precision of the integers that is safe from overflows during global sums. This will be larger than the compile-time precision parameter, and is used to detect overflows. overflow Returns true if the conversion is being done on a value that is too large to be represented i_pr mom_error_handler::mom_error nan_error ni pr prec real_to_efp reproducing_efp_sum_2d reproducing_sum_2d reproducing_sum_3d real function real function mom_coms::ints_to_real (ints) ints_to_real ints ints Convert the array of integers that constitute an extended-fixed-point representation into a real number. ints The array of EFP integers ni pr efp_to_real reproducing_sum_2d reproducing_sum_3d subroutine subroutine mom_coms::increment_ints (int_sum, int2, prec_error) increment_ints int_sum int_sum int2 int2 prec_error prec_error Increment an array of integers that constitutes an extended-fixed-point representation with a another EFP number. int_sum The array of EFP integers being incremented int2 The array of EFP integers being added prec_error The PE-count dependent precision of the integers that is safe from overflows during global sums. This will be larger than the compile-time precision parameter, and is used to detect overflows. ni overflow_error prec efp_minus efp_plus reproducing_efp_sum_2d reproducing_sum_3d subroutine subroutine mom_coms::increment_ints_faster (int_sum, r, max_mag_term) increment_ints_faster int_sum int_sum r r max_mag_term max_mag_term Increment an EFP number with a real number without doing any carrying of of overflows and using only minimal error checking. int_sum The array of EFP integers being incremented r The real number being added. max_mag_term A running maximum magnitude of the r's. i_pr max_efp_float nan_error ni overflow_error pr reproducing_efp_sum_2d reproducing_sum_3d subroutine subroutine mom_coms::carry_overflow (int_sum, prec_error) carry_overflow int_sum int_sum prec_error prec_error This subroutine handles carrying of the overflow. int_sum The array of EFP integers being modified by carries, but without changing value. prec_error The PE-count dependent precision of the integers that is safe from overflows during global sums. This will be larger than the compile-time precision parameter, and is used to detect overflows. i_prec ni overflow_error prec efp_list_sum_across_pes efp_val_sum_across_pes reproducing_efp_sum_2d reproducing_sum_3d subroutine subroutine mom_coms::regularize_ints (int_sum) regularize_ints int_sum int_sum This subroutine carries the overflow, and then makes sure that all integers are of the same sign as the overall value. int_sum The array of integers being modified to take a i_prec ni prec efp_to_real reproducing_efp_sum_2d reproducing_sum_3d logical function, public logical function, public mom_coms::query_efp_overflow_error () query_efp_overflow_error Returns the status of the module's error flag. overflow_error mom_spatial_means::global_i_mean mom_spatial_means::global_j_mean subroutine, public subroutine, public mom_coms::reset_efp_overflow_error () reset_efp_overflow_error Reset the module's error flag to false. overflow_error mom_spatial_means::global_i_mean mom_spatial_means::global_j_mean type(efp_type) function, public type(efp_type) function, public mom_coms::efp_plus (EFP1, EFP2) efp_plus EFP1 EFP1 EFP2 EFP2 Add two extended-fixed-point numbers. The result in extended fixed point format efp1 The first extended fixed point number efp2 The second extended fixed point number increment_ints type(efp_type) function, public type(efp_type) function, public mom_coms::efp_minus (EFP1, EFP2) efp_minus EFP1 EFP1 EFP2 EFP2 Subract one extended-fixed-point number from another. The result in extended fixed point format efp1 The first extended fixed point number efp2 The extended fixed point number being subtracted from the first extended fixed point number increment_ints ni subroutine subroutine mom_coms::efp_assign (EFP1, EFP2) efp_assign EFP1 EFP1 EFP2 EFP2 Copy one extended-fixed-point number into another. efp1 The recipient extended fixed point number efp2 The source extended fixed point number ni real function, public real function, public mom_coms::efp_to_real (EFP1) efp_to_real EFP1 EFP1 Return the real number that an extended-fixed-point number corresponds with. efp1 The extended fixed point number being converted ints_to_real regularize_ints efp_list_sum_across_pes efp_real_diff efp_val_sum_across_pes real function, public real function, public mom_coms::efp_real_diff (EFP1, EFP2) efp_real_diff EFP1 EFP1 EFP2 EFP2 Take the difference between two extended-fixed-point numbers (EFP1 - EFP2) and return the result as a real number. efp1 The first extended fixed point number efp2 The extended fixed point number being subtracted from the first extended fixed point number The real result efp_to_real type(efp_type) function, public type(efp_type) function, public mom_coms::real_to_efp (val, overflow) real_to_efp val val overflow overflow Return the extended-fixed-point number that a real number corresponds with. val The real number being converted overflow Returns true if the conversion is being done on a value that is too large to be represented mom_error_handler::mom_error real_to_ints mom_energetic_pbl::energetic_pbl_init mom_energetic_pbl::epbl_column subroutine, public subroutine, public mom_coms::efp_list_sum_across_pes (EFPs, nval, errors) efp_list_sum_across_pes EFPs EFPs nval nval errors errors This subroutine does a sum across PEs of a list of EFP variables, returning the sums in place, with all overflows carried. efps The list of extended fixed point numbers nval The number of values being summed. errors A list of error flags for each sum carry_overflow efp_to_real max_count_prec mom_error_handler::mom_error ni overflow_error subroutine subroutine mom_coms::efp_val_sum_across_pes (EFP, error) efp_val_sum_across_pes EFP EFP error error This subroutine does a sum across PEs of an EFP variable, returning the sums in place, with all overflows carried. efp The extended fixed point numbers being summed across PEs. error An error flag for this sum carry_overflow efp_to_real max_count_prec mom_error_handler::mom_error ni overflow_error subroutine, public subroutine, public mom_coms::mom_infra_end mom_infra_end This subroutine carries out all of the calls required to close out the infrastructure cleanly. This should only be called in ocean-only runs, as the coupler takes care of this in coupled runs. Interfaces to non-domain-oriented communication subroutines, including the MOM6 reproducing sums facility.